Summary
jailbreaking-general-purpose-robots
We show that state of the art LLMs can be jailbroken by adversarial multimodal inputs, and that this can lead to dangerous scenarios if these LLMs are used as planners in robotics. We propose finetuning small multimodal language models to act as guardrails in the robot's planning pipeline.
Cite this work:
@misc {
title={
Jailbreaking general purpose robots
},
author={
Axel Backlund, Lukas Petersson
},
date={
9/1/24
},
organization={Apart Research},
note={Research submission to the research sprint hosted by Apart.},
howpublished={https://apartresearch.com}
}